trend models - university of wisconsin–madisonbhansen/390/2010/390lecture6.pdf · trend models...
TRANSCRIPT
![Page 1: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/1.jpg)
Trend Models
• A trend model is
where Timet is the time index.• In STATA, Timet is an integer sequence, normalized to be zero at first observation of 1960.
• Most common models– Linear Trend– Exponential Trend– Quadratic Trend– Trends with Changing SLope
)( tt TimegT =
![Page 2: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/2.jpg)
Warning:Be skeptical of Trend Models
• While in some cases, trend forecasting can be useful.
• In many cases, it can be hazardous.
• We will examine some of the trend examples in Chapter 5 of Diebold’s text
• They did not forecast well out of sample.
• A constructive alternative is to forecast growth rates, as we did for consumption expenditure.
![Page 3: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/3.jpg)
Example 1Labor Force Participation Rate
• From BLS
• Monthly, 1948‐2009, Seasonally adjusted
• Men and Women, ages 25+
• Percentage of population in labor force (employed plus unemployed divided by population)
• Diebold estimates on 1948‐1992
• We will estimate on 1948‐1992, forecast 1993‐2009
![Page 4: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/4.jpg)
Women’s Labor Participation Rate1948‐1992
![Page 5: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/5.jpg)
Men’s Labor Participation Rate1948‐1992
![Page 6: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/6.jpg)
Linear Trend Model
• The labor force participation rates have been smoothly and linearly increasing (for women) and smoothly and linearly decreasing (for men) over 1948‐1992
• This suggests a linear trend
• In this model, β1 is the expected period‐to‐period change in the trend Tt
tt TimeT 10 ββ +=
![Page 7: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/7.jpg)
Example 2Retail Sales, Current Dollars
• From Census Bureau
• Monthly, 1955‐2001, seasonally adjusted– This particular series discontinued after 2001
• Diebold estimates up to 1991
• We will forecast 1992‐current
![Page 8: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/8.jpg)
Retail Sales1955‐1993
![Page 9: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/9.jpg)
Quadratic Trends
• The retail sales series has been increasing smoothly over 1955‐1993, but not linearly.
• To model this we will use a quadratic trend2
210 ttt TimeTimeT βββ ++=
![Page 10: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/10.jpg)
Example 3Transaction Volume, S&P Index
• From Yahoo Finance
• (Similar to NYSE series in Diebold)
• Weekly, 1950‐current
• Diebold estimates on 1955‐1993, forecasts 1994
• We will forecast 1994‐2001
![Page 11: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/11.jpg)
Transaction Volume
![Page 12: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/12.jpg)
Exponential Trend
• To model this we will use an exponential trend
• The exponential trend is linear after taking (natural) logarithms
• This is typically estimated by a linear model after taking logs of the variable to forecast
tTimet eT 10 ββ +=
( ) tt TimeT 10ln ββ +=
![Page 13: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/13.jpg)
Ln(Volume)
• In logarithms, trend is roughly linear.
![Page 14: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/14.jpg)
Exponential Trends
• Most economic series which are growing (aggregate output, such as GDP, investment, consumption) are exponentially increasing– Percentage changes are stable in the long run
• These series cannot be fit by a linear trend
• We can fit a linear trend to their (natural) logarithm
![Page 15: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/15.jpg)
Linear Models
• The linear and quadratic trends are both linear regression models of the form
or
where– x1t = Timet– x2t = Timet2
ttt xx 22110 βββμ ++=
tt x110 ββμ +=
![Page 16: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/16.jpg)
Example 4Real GDP
• From BEA
• Quarterly, 1947‐2009
• We will estimate on 1947‐1990, forecast 1991‐2009
• Also use an exponential trend
![Page 17: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/17.jpg)
Real GDP
![Page 18: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/18.jpg)
Ln(Real GDP)
![Page 19: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/19.jpg)
Linear Forecasting
• The goal is to forecast future observations given a linear function of observables
• In the case of trend estimation, these observables are functions of the time index
• In other cases, they will be other functions of the data
• In the model the forecast for yt+h is ŷt+h=b0+b1xt where b0and b1 are estimates
tt x10 ββμ +=
![Page 20: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/20.jpg)
Estimation
• How should we select b0 and b1 ?• The goal is to produce a forecast with low mean square error (MSE)
• The best linear forecast is the linear function β0+β1xt that minimizes the MSE
• We do not know the MSE, but we can estimate it by a sample average
( ) ( )2102ˆ thththt xyEyyE ββ −−=− +++
![Page 21: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/21.jpg)
Sum of Squared Errors
• Sample estimate of mean square error is the sum of squared errors
• The best linear forecast is the linear function β0+β1xt that minimizes the MSE, or expected sum of squared errors.
• Our sample estimate of the best linear forecast is the linear function which minimizes the (sample) sum of squared errors.
• This is called the least‐squares estimator
( ) ( )∑=
+ −−=n
tthtn xy
nS
1
21010
1, ββββ
![Page 22: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/22.jpg)
Least Squares
• The least‐squares estimates (b0,b1) are the values which minimize the sum of squared errors
• This produces estimates of the best linear predictor – the linear function β0+β1xt that minimizes the MSE
( ) ( )∑=
+ −−=n
tthtn xy
nS
1
21010
1, ββββ
![Page 23: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/23.jpg)
Multiple Regressors
• There are multiple regressors
• For example, the quadratic trend
• The best linear predictor is the linear function β0+β1x1t+β2x2t that minimizes the MSE
2210 ttt TimeTimeT βββ ++=
ttt xx 22110 βββμ ++=
( ) ( )2221102ˆ tthththt xxyEyyE βββ −−−=− +++
![Page 24: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/24.jpg)
Multiple Regression
• The sample estimate of the best linear predictor are the values (b0,b1,b2) which minimize the sum of squared errors
• In STATA, use the regress command
( ) ( )∑=
+ −−−=n
ttthtn xxy
nS
1
222110210
1,, ββββββ
![Page 25: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/25.jpg)
Example 1Women’s Labor Force Participation Rate
![Page 26: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/26.jpg)
Regression Estimation
![Page 27: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/27.jpg)
In‐Sample Fit
![Page 28: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/28.jpg)
Residuals
• Residuals are difference between data and fitted regression line
tht
thtt
TimebbyTye
10
ˆ−−=
−=
+
+
![Page 29: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/29.jpg)
Residual Plot
![Page 30: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/30.jpg)
In‐Sample Fit
• Compute with predict command• Fit looks good
![Page 31: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/31.jpg)
Forecast
• Forecast is the linear function with estimated coefficients
• Compute with predict command
hThT TimebbT ++ += 10
![Page 32: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/32.jpg)
Forecast Intervals
• Compute residuals
• Compute quantiles of residuals– These are constant over time
• Add to predicted values– Identical to constant mean case
tht
thtt
Timebbyye
10
ˆˆ−−=
−=
+
+ μ
![Page 33: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/33.jpg)
Out‐of‐Sample Forecast
• Out of sample prediction might be too low.
![Page 34: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/34.jpg)
Out‐of‐SampleWomen’s Labor Force Participation
• No: Prediction was way too high!
![Page 35: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/35.jpg)
Men’s Labor Force Participation Rate
![Page 36: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/36.jpg)
Estimation
![Page 37: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/37.jpg)
In‐Sample Fit
![Page 38: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/38.jpg)
Residuals
![Page 39: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/39.jpg)
Forecast
• End of Sample looks worrying
![Page 40: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/40.jpg)
Out‐of‐SampleMen’s Labor Force Participation
• Linear Trend Terrible
![Page 41: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/41.jpg)
Example 2Retail Sales
![Page 42: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/42.jpg)
Linear and Quadratic Trend
![Page 43: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/43.jpg)
Linear and Quadratic Trend
![Page 44: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/44.jpg)
Forecast
![Page 45: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/45.jpg)
Residuals
![Page 46: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/46.jpg)
Actual Values
![Page 47: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/47.jpg)
Example 3: Volume
![Page 48: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/48.jpg)
Estimating Logarithmic Trend
![Page 49: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/49.jpg)
Fitted Trend
![Page 50: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/50.jpg)
Residuals
![Page 51: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/51.jpg)
Forecast
![Page 52: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/52.jpg)
Out‐of‐Sample
![Page 53: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/53.jpg)
Forecasting Levels from a Forecast of Logs
• Let Yt be a series and yt=ln(Yt) its logarithm• Suppose the forecast for the log is a linear trend: E(yt+h | Ωt) = Tt = β0+ β1 Timet
• Then a forecast for Yt is exp(Tt)• If [LT , UT] is a forecast interval for yT+h• Then [exp(LT) , exp(UT)] is a forecast interval for YT+h
• In other words, just take your point and interval forecasts, and apply the exponential function.– In STATA, use generate command
![Page 54: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/54.jpg)
Forecast in Levels
![Page 55: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/55.jpg)
Out‐of‐Sample
![Page 56: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/56.jpg)
Example 4: Real GDP
![Page 57: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/57.jpg)
Ln(Real GDP)
![Page 58: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/58.jpg)
Estimation
![Page 59: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/59.jpg)
Fitted Trend
![Page 60: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/60.jpg)
Residuals
![Page 61: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/61.jpg)
Forecast of ln(RGDP)
![Page 62: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/62.jpg)
Forecast of RGDP (in levels)
![Page 63: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/63.jpg)
Out‐of‐Sample
![Page 64: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/64.jpg)
Problems with Pure Trend Forecasts
• Trend forecasts understate uncertainty• Actual uncertainty increases at long forecast horizons.
• Short‐term trend forecasts can be quite poor unless trend lined up correctly
• Long‐term trend forecasts are typically quite poor, as trends change over long time periods
• It is preferred to work with growth rates, and reconstruct levels from forecasted growth rates (more on this later).
![Page 65: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/65.jpg)
Trend Models
• I hope I’ve convinced you to be skeptical of trend‐based forecasting.
• The problem is that there is no economic theory for constant trends, and “changes” in the trend function are not apparent before they occur.
• It is better to forecast growth rates, and build levels from growth.
![Page 66: Trend Models - University of Wisconsin–Madisonbhansen/390/2010/390Lecture6.pdf · Trend Models • A trend model is where Time t. is the time index. • In STATA, Time. t. is an](https://reader034.vdocument.in/reader034/viewer/2022042115/5e91ca3000aed6088e6d7ddc/html5/thumbnails/66.jpg)
Final Trend ForecastWorld Record – 100 meter sprint